 | |||||||||||
| |||||||||||
 |
|||||||||||
| |
| ||||||||||||||||
| ShaoLin HA Cluster 1.0 Installation Manual and Operations Guide | ||
|---|---|---|
| Prev | Appendix A. Supplementary Software Information | Next |
Global options are stored in the file /etc/shaolin-ha/slha.conf, it contains cluster wide global configurations. The configuration file exists in a key-value pair basis. A line begins with a # means a comment and will be ignored by the cluster software.
The startup timeout value in N seconds for the Cluster Manager. During the system boot, Cluster Manager will wait for the heartbeat messages from other clusters. If no response or no messages within this period, the Cluster Manager will treat the active cluster doesn't exists and takeover the system. Letting the server to boot to a normal active production state. If this value is not given, the default value is 10.
It is the timeout value in seconds which determines the active server is dead. It must be larger than all KEEP_ALIVE values and is default to 10 if not set. If KEEP_ALIVE value is detected to be larger than the specified value. The standby server will takeover the system from the active server.
In case when two clusters boot at the same time, the HA Cluster will have to determine which server is going to be used as the active server. The Cluster Manager compares the MAC address of the servers by using the given BOOT_ORDER .
The master server MAC address. If this is equal to one of the standby server's MAC address, the standby server will takeover the system even the system is working properly. This is used for system that always wanted to associate a particular server for production.
The number of seconds wait for a sync command to finish before reboot or shut down of the active server. A sync command is issued during system failover to flush all dirty buffers to disk. In case you have a hard disk controller problem or OS problem, the sync command may not be able to complete. This value is used as a time limitation of the sync command. If you have a power switch, the standby server will wait for N seconds before power cycle the active server.
The log directory to store log files for ShaoLin HA Clusters. The default is /var/log/shaolin-ha.
Number of logs to keep for system event dumps. When a log file is dumped by the Cluster Manager, the log files will be rotated.
The log level of the Cluster system. The higher the log level, the more log message it will generate.
The global default keep alive cycle in Nmilli seconds for heartbeat channels. If a separate KEEP_ALIVE value is given in individual heartbeat channel configuration file, the global KEEP_ALIVE value will be ignored.
You have to set this to YES to enable remote cluster manager to connect to the Cluster Manager from IP networks. If this value is not given, the default is set to YES. Set to NO to disable remote control.
If you want the standby cluster to able to communicate to other IP networks other than the local area network, you have to specify the gateway IP address. This value is optional.
In case you want to use a dedicated server machine to run your application service, you should specify one of its MAC address of its NIC's here. If a standby server is recognized as a master, even there is an active server running on-line correctly, the standby server (the master server) will issue a takeover request to the standby server and takeover the system.
In the event of a failure, the Cluster Manager needs to reset the cluster. If this value is not given, the default is to use REBOOT.
The SMTP server for sending out email notifications. You must set a valid SMTP server if you want to receive events by email.
The sender email address of the email sending out by the HA Cluster.
The recipient email address of the email sending out by the HA Cluster.
The SNMP server that the SNMP traps will be sent to.
Extra kernel modules you want to load in initrd.
